One of the primary complaints was that an unmonitored bot has the power to override years of work done by volunteer translation teams.

First you have to stop the destruction. Then you can talk about how to make the bot work for humans, instead of against them.

After this, the common ground can only be that they translate something else. Overriding human translations for some crap done with LLM that is obviously inferior... are they ok?

They had specific complaints. Either say, "we're willing to work to change the things that you're complaining about", or say, "sorry, we're not going to change those things". The wishy washy bullshit they did is effectively ignoring the guy. They're ignoring everything the person said. And it definitely comes across as "sorry not sorry" corporate bullshit.

The standard for fixing something this broken is stop the bot, revert the rollout, revert changes to pages affected, then attempt to figure out what went wrong. Even if you haven't done it yet, announce the revert steps, FFS.

Revert the changes and then hop on a call to figure out how to do it right next time.

"Explain what's wrong and how to fix it" is the wrong approach. if you need it explained to you what was broken, then you're not the expert here, just the local tinpot dictator.